Viewing profile — throwa356262

Same thing with the example.com domain until RFC 2606 put an end to it.

This went quickly from free marketing to possible legal liability for providers and customers!

So random script hits their network and suddenly !! free PR !! after some really questionable assumptions and conclusions from their side?

In the mean time, my very real vulnerability reports are not even acknowledged because maintainers are busy handling this kind of nonsense.

Had no idea Larry was pushing to become the new Murdoch.

Did you see this section? To the B200’s defence, its numbers are somewhat deflated by the fact that it pays a cross-node all-reduce on the decode critical path (RoCE v2 at ~195 Gb/…

Elon-ification begins...

Half of the engineers at OAI and Anthropic are Asian, I don't think China does all these for signaling.

Huawei A910 is supposed to have native support for MXFP4. The problem is that huawei doesn't have the capacity to build more GPUs right now. Those yield numbers must be really terr…

Gentlemen, start your DGX Sparks https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731

What harness are you using to achieve that level of token caching?

So you are telling me Irregular has (a) a copy of Mythos 5 and (b) bad internal security? I bet these guys will receive some sophisticated phishing emails before the day has ended.…

If the hardware is good and cheap, it should be a fun project to replace the OS with a custom Android build that is clean of adware. Do you have a link to the projector?

Good idea, you got my vote! ... unfortunately I don't have a vote according to Tobi https://fortune.com/2026/07/27/shopify-ceo-voting-rights-str...

Could still be 80% marketing. These models are trained on cyber intrusion, that's literally what ExploitGym benchmark measures. That part should not surprise anyone. But what if, s…

Cheng Chye Hsern [...] said Apple is one the few tech firms "not taking part in the AI race" They certainly did take part in the AI race. It's just that their public attempts so fa…

More like RNN. The nvidia version is heavy to compute (common problem with RNN and LSTM, also noted in the paper). Moonshot's Kimi K3 replaces part of it with some function that pe…

Same with DeepSeek papers, they are a joy to read.

For K3 read this instead: https://arxiv.org/abs/2607.24653 The main contribution of the K3 paper is Stable LatentMoE. Like some other models it compresses data sent between layers,…

Yeah, it looks like he mixed up the responses from the study group and this resulted in multiple errors in the published version. But it is also hard to take this seriously when ma…

"Opus 5 states the hidden rules before its first action, then plays a byte-identical optimal solution in 5/5 seeds at temperature 1.0. Zero exploration." I guess there is no way th…

Ah, miss that one. But it still looks incomplete: "Due to the specifics of Kimi K3’s hosting setup, UK AISI / CAISI ran a selective set of cyber evaluations." ... and "Kimi K3’s ov…

Is NIST referring to a yet to be published AISI report? The latest public AISI report says they are waiting with K3 evaluation until weights are published. Am I reading this wrong?…